Abstract:
A computational graceful degradation method is disclosed using priority information in multiple objects case. The method degrades the amount of computation in different ratio every object using the priority information if computational power of a decoder does not come up to that necessary for processing encoded multimedia data as decoding the multimedia data encoded by several objects unit, the priority information being determined by an encoder for optimal graceful degradation in terms of relatively importance between respective objects.
Abstract:
Un método de degradación suavizada de cálculo que se ha de implementar en un sistema codificador y decodificador MPEG, que comprende: un codificador que incluye una unidad divisora de objetos multimedia para dividir una entrada multimedia en una pluralidad de objetos según su contenido y característica; una pluralidad de unidades codificadoras de objetos para codificar los correspondientes objetos multimedia de entrada procedentes de la unidad divisora de objetos multimedia; un multiplexor para multiplexar una pluralidad de corrientes de bits de objetos multimedia codificados para transmitir una corriente de bits de objetos multimedia multiplexada; y un decodificador que incluye un desmultiplexor para desmultiplexar la corriente bits de objetos multimedia multiplexada, recibida del multiplexor, para dar salida a una pluralidad de corrientes de bits de objetos multimedia desmultiplexada; una pluralidad de unidades decodificadoras de objetos para decodificar la correspondiente corriente de bits de objetos multimedia aplicada desde el desmultiplexor; y una unidad de construcción de objetos multimedia para reconstruir los objetos multimedia decodificados a fin de generar objetos multimedia reconstruidos para el usuario; comprendiendo el método las etapas de: determinar, en el multiplexor, una prioridad de los respectivos objetos según la importancia relativa entre los objetos, antes de multiplexar las corrientes de bits de objetos codificados a transmitir, considerando que la degradación suavizada de cálculo se aplicará en el decodificador; y determinar, en el desmultiplexor, una cantidad de cálculo que hay que degradar en proporción diferente cada objeto de vídeo, usando la información de prioridad determinada según los respectivos objetos.
Abstract:
A computational graceful degradation method is disclosed using priority information in multiple objects case. The method degrades the amount of computation in different ratio every object using the priority information if computational power of a decoder does not come up to that necessary for processing encoded multimedia data as decoding the multimedia data encoded by several objects unit, the priority information being determined by an encoder for optimal graceful degradation in terms of relatively importance between respective objects.
Abstract:
PURPOSE: A method for generating a natural language based on a portable terminal and the portable terminal thereof are provided to generate a natural language corresponding to an action of a user. CONSTITUTION: Action information of a user is collected by a portable terminal(120). The collected information is classified(130). A search object exists in the classified information(140). Information is obtained by searching the search object by a web(150). A natural language corresponding to an action of a user is generated based on the classified information and the obtained information(160).
Abstract:
PURPOSE: A translation performance evaluating method and a device thereof are provided to automatically perform evaluation by a domain, a revision, and a dictionary, thereby increasing translation performance of an automatic translation engine. CONSTITUTION: A generation unit receives and compiles a newly registered engine source and registers a complied execution file. A storage unit(102) receives a first translation knowledge dictionary and registers the first translation knowledge dictionary in a second translation knowledge dictionary. A first evaluation unit(103) translates the execution file based on an evaluation set generated by the second translation knowledge dictionary, a tuning set, and an evaluation set. The tuning set is generated by a manager. The evaluation set is generated by an evaluator.
Abstract:
PURPOSE: A schedule management method and apparatus thereof are provided to automatically give notification for scheduled items by automatically extracting schedule information from an e-mail or voice of a user. CONSTITUTION: When the voice of a user is inputted, the voice is changed into text by recognizing the voice(110, 111). Language is analyzed for the changed text or e-mail. Schedule establishment information is extracted from a time processing result(120). The established schedule is notified to the user(160). The established schedule is controlled based on feedback information.
Abstract:
PURPOSE: A method for voice recognition for performing the voice recognition about the noises is provided to differentially change the generation degree by division of the signal according to reliability information. CONSTITUTION: An interval classifying unit(110) compartmentalizes the inputted voice signal into the plural periods of the specific unit. A Fineness determining unit(120) determines the granularity of the grid generation among plural periods about one section. A grating generating unit(130) generates a lattice for the voice recognition according to granularity. A decoding unit(140) acquires a voice recognition result by performing the search space.